Live events / Shows

Live shows — concerts, awards, talk shows

Segment-aware analytics for any live event that isn't a sports match. Audience engagement is first-class — votes, polls, reactions, chat, donations, super-chat.

How shows differ from sports

Sports have phases and a score. Shows have segments and engagement. The dashboard renders these differently:

A wicket or goal becomes a marker on the timeline. A segment change becomes a segment_start marker plus a new bar on the segment-retention chart.
Score progression matters in sports. It doesn’t in shows — what matters is which segments held the audience and which lost them.
Audience reactions are passive in sports. In shows they’re active — viewers vote, react, donate, super-chat. We capture those as first-class engagement events, not just generic markers.

Concept: the active show context

Calling startLiveShow(ctx) on the SDK installs a show context that the SDK stamps onto every event you emit until you call stopLiveShow(). This works just like the live-match context but with show-flavored fields — and it auto-stamps the current live_segment and live_segment_index so retention queries can correlate every event with which segment was on screen.

typescript

type LiveShowContext = {
  eventId: string;        // = the match_id you registered + content_id you send
  segment?: string;       // free-form segment name e.g. 'Best Actor'
  segmentIndex?: number;  // zero-based ordering, useful for sorting
  getLatencyMs?: () => number | undefined; // sampled per event
};

Concert example

Concerts are mostly performance-driven. Use segments to mark each artist or set, and reactions to gauge audience response in real-time. Donations are common for ticketed virtual concerts.

typescript

import Tapemetric from '@tapemetric/analytics';

const tm = Tapemetric.init({ apiKey: 'tm_live_yourkey' });

// Viewer joins
tm.startLiveShow({
  eventId: 'arijit_singh_live_2026',
  segment: 'pre_show',
  segmentIndex: 0,
  getLatencyMs: () => Math.round((hls.liveSyncPosition - video.currentTime) * 1000),
});

tm.trackPlayStart({
  contentId: 'arijit_singh_live_2026',
  contentType: 'live',
});

// Set 1 begins — fires segment_start marker, updates auto-stamp
tm.setSegment('Set 1 — Tum Hi Ho medley', 1);

// Audience reactions
tm.trackReaction('heart');
tm.trackReaction('fire');

// A super-chat tip during the encore
tm.setSegment('Encore — Channa Mereya', 5);
tm.trackDonation(500, { display_name: 'Riya', message_length: 24 });

tm.stopLiveShow();

Awards show example

typescript

tm.startLiveShow({
  eventId: 'filmfare_2026',
  segment: 'red_carpet',
  segmentIndex: 0,
});

tm.trackPlayStart({ contentId: 'filmfare_2026', contentType: 'live' });

// Hosts come on
tm.setSegment('Opening monologue', 1);

// People's choice voting opens
tm.trackShowMarker('vote_open', 'Best film — public vote');
tm.trackPoll('best_film_2026', 'Animal');
tm.trackPoll('best_film_2026', 'Jawan');
tm.trackShowMarker('vote_close', 'Best film — public vote');

// Best Actor segment
tm.setSegment('Best Actor', 4);
tm.trackShowMarker('winner_announced', 'Best Actor — Shah Rukh Khan');
tm.trackReaction('clap');

// In Memoriam
tm.setSegment('In Memoriam', 8);

tm.stopLiveShow();

Talk show with live polls

typescript

tm.startLiveShow({
  eventId: 'koffee_with_karan_s8e12',
  segment: 'opening',
  segmentIndex: 0,
});

// Move to Rapid Fire round
tm.setSegment('Rapid Fire', 3);

// Chat metadata only — never the message body
tm.trackChatMessage({ length: 42, has_emoji: true });

// Audience poll embedded in player
tm.trackPoll('best_guest_appearance', 'Deepika Padukone');

// Vote for next week's guest
tm.trackVote('Ranveer Singh', { context: 'next_week_guest' });

tm.stopLiveShow();

iOS

swift

import TapemetricAnalytics

Tapemetric.shared.startLiveShow(
    eventId: "filmfare_2026",
    segment: "red_carpet",
    segmentIndex: 0,
    getLatencyMs: { [weak self] in
        guard let p = self?.player else { return nil }
        return Int(p.liveLatencySeconds * 1000)
    }
)

Tapemetric.shared.setSegment("Best Actor", index: 4)

Tapemetric.shared.trackShowMarker(
    kind: "winner_announced",
    label: "Best Actor — Shah Rukh Khan"
)

Tapemetric.shared.trackReaction(kind: "clap")
Tapemetric.shared.trackDonation(amountInr: 500.0)

Tapemetric.shared.stopLiveShow()

Android

kotlin

import com.tapemetric.analytics.Tapemetric

Tapemetric.startLiveShow(
    eventId = "koffee_with_karan_s8e12",
    segment = "opening",
    segmentIndex = 0
)

Tapemetric.setSegment("Rapid Fire", index = 3)

Tapemetric.trackPoll("best_guest_appearance", "Deepika Padukone")
Tapemetric.trackReaction("fire")
Tapemetric.trackChatMessage(mapOf("length" to 42, "has_emoji" to true))

Tapemetric.stopLiveShow()

Engagement event reference

Method	Event type	Use it for
`trackVote(option)`	`vote`	Single-pick votes — best song, fan favourite, next-week guest
`trackPoll(pollId, choice)`	`poll_response`	Multi-question polls during a show
`trackReaction(kind)`	`reaction`	Emoji reactions during performances
`trackChatMessage(props)`	`chat_message`	Chat metadata only (length, has_emoji) — not the body
`trackDonation(amount)`	`donation`	Tips / super-chat with INR amount
`trackShowMarker(kind, label)`	`live_marker`	Custom moments — winner announced, encore, host change

Privacy note. Don’t send the chat message body to trackChatMessage. Send length and emoji presence as properties; the message text has no analytics value and creates a privacy problem.

Segment retention algorithm

The retention chart is computed by collecting all distinct anonymous_id values per segment, then taking the set intersection with the previous segment to compute "held_from_previous". Hold rate is intersection size / previous segment size. This is more meaningful than a sliding-window concurrency view because it captures viewer-level continuity — a viewer who left at the In Memoriam segment and came back for Best Picture counts as having dropped, not held.

Segments endpoint

bash

GET /v1/live/matches/{id}/segments
Authorization: Bearer eyJ...

200 OK
{
  "match_id": "filmfare_2026",
  "title": "Filmfare Awards 2026",
  "kind": "awards",
  "segments": [
    { "index": 0, "label": "red_carpet",
      "viewers": 250000, "held_from_previous": null, "hold_rate": null },
    { "index": 1, "label": "Opening monologue",
      "viewers": 410000, "held_from_previous": 230000, "hold_rate": 0.92 },
    { "index": 4, "label": "Best Actor",
      "viewers": 580000, "held_from_previous": 380000, "hold_rate": 0.95 },
    { "index": 8, "label": "In Memoriam",
      "viewers": 380000, "held_from_previous": 320000, "hold_rate": 0.65 }
  ]
}

Engagement endpoint

bash

GET /v1/live/matches/{id}/engagement
Authorization: Bearer eyJ...

200 OK
{
  "match_id": "filmfare_2026",
  "title": "Filmfare Awards 2026",
  "kind": "awards",
  "engagement": {
    "vote":          { "count": 47210,  "unique_users": 32104 },
    "poll_response": { "count": 18933,  "unique_users": 14889 },
    "reaction":      { "count": 142550, "unique_users": 38120 },
    "chat_message":  { "count": 23005,  "unique_users": 9180  },
    "donation":      { "count": 412,    "unique_users": 387   }
  }
}

Operational best practices

Plan segments before broadcast. A rough rundown with 10–20 segments works well. Too few and the retention curve is uninformative; too many and the dashboard gets noisy.
Use stable segment names. 'Best Actor' not 'Best Actor (40 min in)'. The dashboard groups by exact label.
Throttle reactions. If your UI lets viewers tap a heart 10 times per second, throttle to one per second on the client. Otherwise a hundred-thousand viewer concert generates a million events per second of reactions.
Pair vote_open / vote_close markers with the actual voting window so you can compute participation rate (votes / unique viewers during the window).
For watch parties, set kind: 'watch_party' and use segments for the underlying content’s natural breaks (episode 1, episode 2). The host’s commentary sits in 'host_change' markers.

When to use match vs show kind

If the event has teams and a score, use kind: 'match'. If it has segments and audience interaction, use kind: 'show' (or one of the more specific kinds). For hybrid events — say a cricket exhibition with celebrity halftime performances — pick whichever metric matters more for your business and use markers/segments to capture the rest. The dashboard renders one or the other; it doesn’t switch mid-event.